A Project is Born

9 February 2011

Schematic of a Wireless Garden MonitorI decided to take a personal day today, and I sat down and knocked out the schematic for a project I’ve wanted to work on for a very long time.

This is a  wireless mesh-networking garden monitor.  It’s configured to monitor soil moisture, soil temperature, ground temperature, as well as air temperature and humidity.

This is interesting information to know, as it provides significant insight into microclimates on a given plot.  It can also reveal information about how well your soil retains moisture, etc. etc.

It will be powered by solar cells, which will opportunistically charge a LiPo battery.  The wireless communication is provided by an Xbee, which is mounted on the back.  The microcontroller is an Atmega 328p.  A Microchip MCP73833 charges the LiPo whenever the solar voltage is adequate, and a Micrel 5205 regulates this voltage.  The device is configured for reprogrammability over the Xbee link.  I’ll probably also add an FTDI cable port for easy debugging early in the development process.

This portion of the board will sit near the ground.  There is a below ground and above-ground portion, as well.  The above ground portion contains a humidity sensor, temperature sensor, and 1W solar cell.  The below ground portion will contain a soil moisture sensor and temperature sensor.  I’m going to cover the whole thing in conformal coating, and see how long it lasts in the elements.

The whole setup will likely be mounted on a 5 or 6 foot length of PVC pipe.  In order to get decent air temperatures, you have to get a few feet off the ground.  NOAA puts their temperature monitors 2M above ground (I think).  NOAA also mounts their temperature sensors in a baffled cage to avoid complications from direct solar heating.  Fortunately, Tim made a printable thermal shield for the White Star project which will work just great! (http://www.thingiverse.com/thing:5725).

At this point, I’m not sure if humidity will be useful data to gather, so I may leave it off the final devices.

The below ground portion will be a dead-bugged temperature sensor and a soil moisture sensor made from plaster of Paris: http://www.cheapvegetablegardener.com/2009/11/how-to-make-cheap-soil-moisture-sensor-2.html

The 1 watt solar cell should be more than enough power for this thing.  As long as the average sun incidence is greater than 2 sun-hours per day, everything will work out great.  Here is the power usage breakdown:

Atmega 328: 4 ma Active, 1 ua Sleep
HH10D Humidity: .2 ma
TMP100: .01 ma
Xbee: 50ma active,  .01 ma Sleep
The voltage regulators and the battery charge circuitry should have nearly negligible quiescent draws.  Assuming 5 seconds of Active time every 60 seconds, we have:
.2+(.01*3)+ (4*5/60)+(.001*55/60)+(50*5/60)+(.01*55/60) = 5.6 mA average power draw at 3.3v.  This is 445 mW-H per day.  Kentucky gets an average of 4 sun-hours per day, so a 200 mW panel should be more than enough.  To keep things on the safe side, though, a 1W panel with a 3000 mW LiPo should be more than enough keep this thing going day and night, with 1 minute resolution of data.
This board is pretty hastily laid out, so I might redo it before risking a manufacturing run.  It shouldn’t matter a whole lot, though, since these are all low speed signals in a pretty forgiving environment.  Most concerning are the proximity of some traces, as well as overlapping signal traces.  I2C bus length may become an issue as well, but I have no qualms about slowing the bus speed down as low as possible.